A study of early predictors (Comorbidities and Etiology) of patient’s outcome after cirrhosis hospitalization

Group 19

Eva Frossard, Pauline Charpentier, Noy Tabul, Fabian Ziegler

Introduction

Objective: Understand patterns and correlations between early predictors Comorbidities and Etiologies, their numerical estimate CPS and Charlson index and cirrhosis’ patients outcome  

Cirrhosis = condition in which the liver is scarred and permanently damaged.

Relevancy of the study: Cirrhosis is a leading cause of mortality in the world (11th), and finding accurate descriptive index is a useful tool to investigate the possible outcome of patients.

Materials - Dataset

Data set used: Early predictors of outcomes of hospitalization for cirrhosis and assessment of the impact of race and ethnicity at safety-net hospitals

  • 733 patients
  • From 4 safety-net hospitals in the US
  • Male dominated study (67.31%)
  • Predominant age group [60; 70]
  • Main liver disease diagnosis
    • Ascites (31%)
    • Hepatic encephalopathy (21.2%)
    • UGIB (19.9%)

Methods

Cleaning:

  • Extracting, renaming, reordering columns
  • Dropping non-available values them

Tidying:

  • Pivot_wider on Comorbidities
  • Pivot_wider on Etiologies

Augmenting:

  • Created categories for Mortality: no death, death in hospital, after 30 days, after 90 days
  • Creating age bins instead decade

Overview of comorbidities and etiologies

  • Etiology is the study of the factors that come together leading to a disease.

  • Comorbidity is an additional disease that can interact and coexist simultaneously with cirrhosis.

Results: Different outcomes tendency depending on the liver disease

  • Bar plot which shows liver disease distribution, filled by mortality status

  • Demonstrates the impact of varying mortality rates base on liver diseases

  • Reveals differences in mortality timing between “Other fluid overload” and “SNB”

Death ratios of every Etiology/Liver disease combination

  • Tile chart reveals ratios for liver diseases and associate etiologies

  • High mortality for specific combinations like Hepatic encephalopathy/Auto-immune hepatitis.

  • Lower death ratios (<25%) for example for Hepatitis B or Non-alcoholic fatty liver disease.

  • Over representation of certain comorbidities

Model 1 - Logistic regression

  • Overview: Logistic regression model

  • Variables: CPS & Mortality

  • Results: For every increased CPS unit, the likelihood for mortality increase by 50 %

Model 2 - Linear regression

  • Overview: Linear regression model

  • Variables: CPS & Etiologies

  • Results: Statistically significant result for nafld associated with low CPS

Conclusion

  • The type of the liver disease notably impacts the patient’s outcome

  • The difference in timing (less or more fulgurant) hint a varying disease severity

  • Lethal combinations of liver diseases and their associated etiologies can be identified

  • For every CPS unit increase there is a 50 % higher likelihood for mortality

  • According to the linear regression nafld is associated with low CPS